You are on page 1of 73

# Coupled Analysis of FSI

CCFD

## Important Key Words

Fluid Structure Interface = FSI Computational Fluid Dynamics = CFD Pressure Displacement Analysis = PDA Thermal Stress Analysis = TSA

CCFD

## A Part of the Reference

CCFD

Why It Is Important ?
Automobile (export, engine) Construction (bridge, road etc.) Electronic parts (base, MEMS) anywhere you can meet the needs of the coupled analysis of thermal flow and thermal stress.
CCFD
4

## What Kind of Analysis Should be ?

We can divide them into 2 types
Pressure on the solid surface
Fluid and pressure displacement of the solid

## Convection coefficient on the solid surface

Fluid and thermal stress of the solid

CCFD

## Pressure on The Solid Surface

Small displacement ( Linear ) Big displacement
Linear ,Nonlinear

Transient
Small displacement ( Linear ) Big displacement
Linear ,Nonlinear

Fatigue of displacement
CCFD
6

CCFD

CCFD

Transient FSI

CCFD

Transient FSI

CCFD

10

Transient FSI

CCFD

11

## Convection on The Solid Surface

Temperature distribution in the solid part and the thermal stress must be considered
Linear of Property Nonlinear of Property

## Fatigue of Thermal Stress

CCFD

12

The Status
Always define a uniform film coefficient at the solid part surface

CCFD

13

The Problems
Effect of the flow can not be estimated correctly, especially for turbulent flow

CCFD

14

The Problems
This kind of analysis is not easy to do
You must understand CFD (computational fluid dynamics) You must be familiar with thermal stress analysis (TSA) too

CCFD

15

The Problems
When problem spans both TSA and CFD, the analysis procedure may be less established. This typically occurs because:
a stress analysis system and a CFD system tend to be developed independently from each other much less attention is paid to the interface between them, although both of CFD and TSA are classic subjects.

CCFD

16

## How to Solve the Problems

Experiment: To measure the film coefficientcan define a reasonable film coefficient, but it takes cost and time, and the measure points are limited
Too difficult or impossible

CCFD

17

## How to Solve the Problems(2)

Calculation(1): to obtain a converged solution. Inner and outer thermal flow around (not including ) the solid part are simulated to get the correct film coefficient for thermal and thermal stress analysis of solid part 3 steps : thermal flow analysis (flow part ) thermal analysis (solid part ) thermal stress analysis (solid part)
CCFD
18
A few iterations between stages might be necessary

## How to Solve the Problems(3)

Calculation(2): Inner and outer thermal flow around ( including ) the solid part are simulated to get the correct solid temperature for thermal stress analysis directly 2 steps : thermal flow analysis (solid +flow) thermal stress analysis (solid)
CCFD
19

## The Possible Choices of CFD+TSA

FVM CFD calculation > FEM structure analysis
FVM > FEM

## FEM CFD calculation > FEM structure analysis

FEM > FEM
FVM (finite volume method) FEM (finite element method) Much easier than experiments

CCFD

20

## Investigation And Comparison

FVM > FEM: SC/Tetra + ANSYS FEM > FEM: FLOTRAN + ANSYS Are used for our investigation and comparison In our investigation, we have used one CFD system, SC/Tetra developed by authors [1]. To solve for the thermal stress field, we have used a commercial structure system ANSYS developed by ANSYS.Inc. Both two applications are capable of solving for temperature fields.
21

CCFD

## Investigation And Comparison

FVM > FEM: SC/Tetra + ANSYS FEM > FEM: FLOTRAN + ANSYS Are used for our investigation and comparison These solvers are chosen because they are readily available to the authors. But authors insist that the concepts introduced here can be used for coupling between any CFD and any structure applications developed by oneself or of commercial software.

CCFD

22

SC/Tetra
supports hexahedral, prismatic, pyramidal and tetrahedral elements, i.e. a hybrid mesh, generating a tetrahedral mesh first and inserting prism layers along no-slip walls. all variables are defined on nodes of elements.
can be used easily to make a complex mesh model and has a good interface with FEM structure analysis software.
CCFD
23

SC/Tetra

generating a tetrahedral mesh first and inserting prism layers along no-slip walls. CCFD
24

FLOTRAN + ANSYS
Various element types for convenient +speed, 72 for convenient + accuracy, 92 for accuracy, 95 45 must be paid attention There is not any limit for the mesh shape,But FLOTRAN can not be easily used for a complex model
CCFD
25

## ICEM and AI-Environment

supports hybrid mesh too generating a tetrahedral mesh first and inserting prism layers along no-slip walls just in the fluid side
ICEM or AI-Environment will help us too

CCFD

26

Element Type

CCFD
27

## Element Type (first order)

first order tetrahedral 4 node element, ET 72 in ANSYS

CCFD

28

## Element Type (second order)

quadratic displacement tetrahedral 10 node element ET 92 in ANSYS

## degenerate quadratic displacement tetrahedral 20 node element ET 95 in ANSYS

CCFD

29

Element Type
Different element types 72 45 92 95 are used in the thermal stress analysis for comparison of element type effect.
CCFD
30

Element Type
CFD analysis can contain hybrid elements in the solid part. In TSA second order degenerate hybrid (such as element type 95) must be considered too. This is a typical difficult of coupling between a CFD system and a structure system for Direct Conversion.

CCFD

31

Other Software
The possible output element shape

## ANSYS DesignSpace NASTRAN I-DEAS ABAQUS

The names of system and products are trade marks or registered marks belong to each company respectively.

CCFD

32

SC/Tetra

Hybrid elements only in the fluid part Only tetra in solid part the coupling process make Direct Conversion very simple and easy. CCFD
33

Typical Model

CCFD

34

Calculation conditions
The inner diameter of the main pipe is 2.0m. The inner diameter of the branch pipe is 1.4m. The thickness of the pipe is 0.1m. The inlet velocity is 0.7 m/s. The inlet temperature is773 K. The velocity of inlet 2 is 0.4 m/s. The temperature of inlet 2 is 573 K. The pressure is 0 Pa at the outlet.
CCFD
35

Calculation conditions(2)
The working fluid is incompressible air density is 1.2050 kg/m3 viscosity is 1.8135e-05 kg/ms thermal conductivity is 0.02574 W/mK specific heat is 1004.0 J/KgK

CCFD

36

Calculation conditions(3)
The material of the pipe is iron. density is 7871.40 kg/m3. Youngs modulus is 210e9 N/m. Poissons ratio is 0.3. thermal conductivity is 81.168 W/mK. specific heat is 439.2 J/KgK. coefficient of linear thermal expansion is 16.6e-6 K-1.
CCFD
37

4 approaches
CFD Delivered Data TSA 0) FLOTRAN(ET142) ANSYS(ET95) T to the same nodes 1) SC/Tetra 2) SC/Tetra 3) SC/Tetra

## T to nodes H mapped to surface T mapped to nodes

CCFD

38

4 approaches(0)
FLOTRAN(ET142) ANSYS(ET95) T to the same nodes

39

4 approaches(0)

## No boundary layer mesh in flow domain in FLOTRAN model.

CCFD
40

4 approaches(1)

CCFD

41

4 approaches(1)
SC/Tetra ANSYS(ET95,72,92) T to the same nodes

Same nodes and same result temperature on solid mesh nodes which are a part of the whole mesh CCFD
42

4 approaches(2)
SC/Tetra ANSYS(ET95,72,92) H is mapped to new surface nodes

## Different nodes and mapped result temperature CCFD

43

4 approaches(3)
SC/Tetra ANSYS(ET95,72,92) T mapped to volume new nodes

## Meshes do not need consistency on the common boundary surface CCFD

44

4 approaches

Direct Conversion FLOTRAN Direct Conversion Surface Mapping SC/Tetra Volume Mapping

Table 2: Number of elements in computational mesh. Both solid and fluid elements are counted in CFD column.

CCFD

45

Results of case 0

## The CFD result of Case 0

Left: velocity field on center cut plane. Right: temperature distribution on the outer surface of the pipe.

CCFD

46

Results of case 1

## The CFD result of Case 1.

Left: velocity field on center cut plane. Right: temperature distribution on the outer surface of the pipe.
47

CCFD

## Temperature distribution on the outer surface of the pipe.

Left: Case 1 Right: Case 0 (obtained using ANSYS).

CCFD

48

## Thermal Stress Comparison(1)

Case 0 ET 95 Case 1 ET 95

CCFD

49

## Conclusion of Transfer Method

The similar results of case 0 and 1 shows that a reasonable accuracy can be obtained using Direct Conversion Method of transferring temperature fields from SC/Tetra to ANSYS as well as FLOTRAN to ANSYS.

CCFD

50

## Thermal Stress Comparison(2)

Case 2 ET 95 Case 3 ET 95

CCFD

51

## Conclusion of Transfer Method

The results in Case 2 and Case 3 are nearly identical and shows that the Surface Mapping and the Volume Mapping can work equally well. The results in all 4 cases are nearly identical and shows that the Direct conversion, Surface mapping and the Volume mapping can work equally well.
52

CCFD

Tmax

Tmax

92
Tmax
CCFD
53

72

## Effect of Element Type look out

45 is typical 3D constant strain triangle element [5] Low order Et 95 will be the same.

45

Tmax
CCFD
54

## Effect of Element Type

Different element types 72 45 92 95 are used in the thermal stress analysis for comparison of element type effect.
CCFD
55

## Effect of Element Type (first order)

first order tetrahedral 4 node element, 72

CCFD

56

## Effect of Element Type (second order)

quadratic displacement tetrahedral 10 node element 92

CCFD

57

## Conclusion of Element Type

Element type 92 gives the closest result to type 95. Element type 45 cannot produce a good result in bending even if a finer mesh is used, because it is a typical 3D constant strain triangle of degenerate tetrahedron Second order elements (type 92 and 95) give better results compared to first order element (type 72).

CCFD

58

CCFD
59

## Using SC/Tetra to do an analysis including solid part

Direct Conversion
Using Solid part mesh and temperature result to do thermal stress analysis ET 95
60

Volume Mapping
Transfer Solid temperature to A finer mesh, and then to do thermal stress analysis ET 45,72,92,95

CCFD

## Example of Exhaust Port

Mesh Model rightfluid domain

CCFD
61

## Finer Mesh 198433(solid)

CCFD
62

Calculation Conditions
mass flow rate at each inlet is 0.0628 kg/s inlet fluid temperature is 1023 K outlet pressure is 138.7 kPa filter condition is applied in the cylinder part film coefficient on the outer wall of the exhaust manifold is set to 2.9 W/m2K and ambient temperature is set to 573 K node movements are fully constrained at two contact surfaces of the flange (last page)
CCFD
63

Calculation Results

air is treated as an ideal gas: Left: Contour plot of Mach number in the filter cylinder center cross section. Right: flow pattern. CCFD
64

Calculation Results

temperature distribution : Left: air is treated as an incompressible fluid. Right: air is treated as an ideal gas. Temperature scale is in centigrade. CCFD
65

Calculation Results

Displacement distribution of exhaust model: Left: Direct conversion Right: Volume mapping CCFD
66

Calculation Results
SC/ Tetra Memory (Mbytes) CPU Time (sec) 25 692 ANSYS Element Type 45 62 699 95 1516 3152 72 92

Table 3: Expected memory and CPU time requirements for a typical computational mesh containing 100,000 elements, CPU time for CFD is the time to advance 100 cycles. The CPU time is estimated using a PC with Pentium III 866 MHz CPU. Note that considerable deviation can result depending on the geometry, mesh quality, shapes of the elements used in the mesh and other factors. CCFD
67

Conclusions
Even Mach Number M<0.7, compressible should be considered Conductivity can be used as constant The Volume Mapping for solid part temperature has priority The priority for choosing element should be 95,92,72

CCFD

68

Conclusions
The combination of SC/Tetra and ANSYS or other TSA is efficient, flexible and accurate. The volume mapping method is the favored method to couple CFD and thermal stress analysis Element 95 low order tetra and 45 tetra should be prohibited

CCFD

69

Conclusions
For Direct Conversion Method, as one mesh is used in 2 stages of the whole analysis, there is no mapping and mapping error. But most thermal stress analyses use only one mesh shape, hybrid mesh is limited to use. In this case, a mesh generation with only tetrahedral in the solid part will make the coupling much easier.
When there is not mesh shape and memory limitation CCFD
70

Conclusions
For Surface Mapping Method: Because it is possible to use a mesh that spans only the fluid part in the CFD stage and to use another mesh that spans only the solid part in the thermal stress analysis stage, there is no limitation of element shape. And this method can be used for a bigger mesh than that in method 1 and 3 with the same resource. In the reverse, a few iterations between stages might be necessary to obtain a converged solution and an accurate mapping is necessary to avoid the mapping error. Especially, the calculations must be divided into 3 stages (thermal flow analysis in flow part, thermal analysis in solid part and thermal stress analysis in solid part), they will take a bit more effort to do.
To divide the solid and fluid part into 2 mesh model to spare memory

CCFD

71

Conclusions
The Volume Mapping Method is most flexible to couple CFD and thermal stress analysis, but is less suitable than Surface Mapping Method for large-scale models.

CCFD

72